Comment by brokensegue

I think better than priors would be a closed loop where you tell it what the right answer was (or some signal) and they monitor and fine-tune for you